From 75c5c2db63de6fc519897318c39d1ebeec0c055b Mon Sep 17 00:00:00 2001
From: Charlotte Som <charlotte@som.codes>
Date: Tue, 25 Feb 2025 17:39:29 +0000
Subject: [PATCH] initial commit

---
 .gitignore          |   2 ++
 README.md           |   7 +++++++
 requirements.txt    | Bin 0 -> 136 bytes
 server/__init__.py  |  11 +++++++++++
 server/http.py      |   5 +++++
 server/inference.py |  33 +++++++++++++++++++++++++++++++++
 6 files changed, 58 insertions(+)
 create mode 100644 .gitignore
 create mode 100644 README.md
 create mode 100644 requirements.txt
 create mode 100644 server/__init__.py
 create mode 100644 server/http.py
 create mode 100644 server/inference.py

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..3f32964
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,2 @@
+/.venv
+__pycache__
diff --git a/README.md b/README.md
new file mode 100644
index 0000000..dbb0ea6
--- /dev/null
+++ b/README.md
@@ -0,0 +1,7 @@
+# llm-ui
+
+```shell
+$ uv venv
+$ uv pip install -r requirements.txt
+$ uv run uvicorn server:app
+```
diff --git a/requirements.txt b/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..620b0c7e037c17e90338a4fd7a840be91da1124f
GIT binary patch
literal 136
zcmX}kOA3H63`Nm97F>!_Y!N3eM#Mp(Dt?!*K8uz&2uV(IUxC0#s~AcuB`0HJ(cH5$
oP;jVv^;>jS&eD8tr(@M_ka_k0$x+Lhs18ou_so)05*C-l18p!BEC2ui

literal 0
HcmV?d00001

diff --git a/server/__init__.py b/server/__init__.py
new file mode 100644
index 0000000..935c59b
--- /dev/null
+++ b/server/__init__.py
@@ -0,0 +1,11 @@
+from server.http import Starlette, Route, Request, Response, JSONResponse, WebSocketRoute
+from server.inference import list_conversations, connect_to_conversation
+
+async def status(request: Request) -> Response:
+  return JSONResponse({"status": "ok"})
+
+app = Starlette(debug=True, routes=[
+  Route("/api/", status),
+  Route("/api/conversation", list_conversations, methods=["GET"]),
+  WebSocketRoute("/api/conversation/:conversation/connect", connect_to_conversation)
+])
diff --git a/server/http.py b/server/http.py
new file mode 100644
index 0000000..e0572b0
--- /dev/null
+++ b/server/http.py
@@ -0,0 +1,5 @@
+from starlette.applications import Starlette
+from starlette.routing import Route, WebSocketRoute
+from starlette.responses import *
+from starlette.requests import *
+from starlette.websockets import WebSocket
diff --git a/server/inference.py b/server/inference.py
new file mode 100644
index 0000000..0c1fd0d
--- /dev/null
+++ b/server/inference.py
@@ -0,0 +1,33 @@
+import llm, llm.cli, sqlite_utils
+from .http import Request, JSONResponse, WebSocket, RedirectResponse
+import json
+
+db = sqlite_utils.Database(llm.cli.logs_db_path())
+
+async def list_conversations(request: Request):
+  conversations = []
+  for row in db["conversations"].rows:
+    conversations.append({ "id": row["id"], "name": row["name"] })
+  return JSONResponse(conversations)
+
+girlypop_prompt = llm.cli.load_template("girlypop").system
+
+async def connect_to_conversation(ws: WebSocket):
+  conversation_id = ws.path_params["conversation"]
+  if conversation_id == "new":
+    conversation = llm.AsyncConversation(llm.get_async_model())
+  else:
+    try:
+      conversation: llm.AsyncConversation = llm.cli.load_conversation(conversation_id, async_=True)
+    except:
+      await ws.send_denial_response(JSONResponse({
+        "error": "unable to load conversation {}".format(conversation_id)
+      }))
+      return
+
+  await ws.accept()
+  async for message in ws.iter_text():
+    response = conversation.prompt(message, system=girlypop_prompt)
+    async for chunk in response:
+      ws.send_text(json.dumps({"c": chunk}))
+    ws.send_text(json.dumps({"d": True})) # done